JP2004030690A - Gateway unit, client computer and proxy server computer - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To effectively use a path by distributing an access request to a file object, to enhance throughput and to further reduce communication quantity on a local network. <P>SOLUTION: A system is constituted of a client computer 24, a server computer 12 which provides information according to an information provision request from the client computer 24, redirector computers 72 to 77 which control the file object of the server computer 12 and proxy server computers 65 to 67 which repeats transfer of the file object based on the information provision request from the client computer 24. The redirector computers 72 to 77 perform control for distributing information quantity to be stored by every proxy server computer by determining a proxy server computer in which requested information is stored based on an operation result by a prescribed arithmetic operation using a character string included in the information provision request and the number of proxy server computers 65 to 67 in the system. <P>COPYRIGHT: (C)2004,JPO

Description

The present invention relates to a distributed file system in which a plurality of server computers distributed on a network and a plurality of client computers are interconnected by a communication line via a gateway device (computer). The present invention relates to a gateway device that relays a file in a distributed file system when accessing a file object in a storage device inside a server computer via a gateway device, a client computer that outputs an access request to the gateway device, and a proxy server computer.

で In the following description, a “server computer” refers to a computer that provides some service on a network, and a “client computer” refers to a computer that requests and receives the services of such a server computer. The “file object” refers to a set of a network protocol used by the distributed file system, a network address (name of a server computer), a file name, and a file entity. Here, the “file object name” refers to a set of a network protocol, a network address, and a file name.

Conventionally, in a distributed file system as described above, a read request for a file object on a server computer from a plurality of client computers has been relayed once by a gateway computer on the way. The file object read from the server computer by the gateway computer is relayed to the client computer. At the same time, these file objects are stored in a cache file (fixed disk, semiconductor memory, etc.) inside the gateway computer. An example of such a gateway computer is disclosed in Japanese Patent Application Laid-Open No. 4-313126 (Title of Invention: "File Input / Output Method of Distributed File System").

FIG. 15 shows this conventional distributed file system. This system includes a gateway computer 110, a server computer 100 and a plurality of client computers 120 interconnected via the gateway computer 110. The server computer 100 is shared by a plurality of client computers 120.

The server computer 100 includes a disk device 136 for storing the file, a file input / output response unit 134 for inputting / outputting the contents of the file in the disk device 136 via the network, and the block information in the disk device 136 via the network. And a block information responding unit 132 for responding to the client computer 120.

The client computer 120 sends a block information request to the block information responding unit 132 of the server computer 100, and a block information requesting unit 152 for receiving block information, and a file input / output response of the server computer 100 via the gateway computer 110. File input / output request means 154 for sending a file input / output request to the means 134 and receiving the file.

The gateway computer 110 is interposed between the client computers 120 (plural) and the server computer 100, and functions as a cache memory between the disk device 136 in the server computer 100 and each client computer 120. And a cache management means 142 interposed between the file input / output request means 154 and the file input / output response means 134.

This system works as follows. First, an operation when the client computer 120 accesses a file in the disk device 136 in the server computer 100 and reads one block will be described. The block information requesting unit 152 issues a block information request message requesting the server computer 100 via the gateway computer 110 to obtain block information on the block to be read.

In response to this message, the block information responding unit 132 in the server computer 100 extracts the corresponding block information from the disk device 136, and sends a block information response message having the block information to the client computer 120 via the gateway computer 110. Is returned to the block information requesting means 152.

(4) The file input / output request unit 154 of the client computer 120 issues a file access request to read the block to be read based on the returned block information to the cache management unit 142 in the gateway computer 110.

(4) The cache management unit 142 that has received the file access request compares the block information relating to the file access request with the block information relating to the block to be read stored in the disk cache 144. When the corresponding block information does not exist in the disk cache 144, or when the contents (update time) of both block information are different (when the cache of the block to be read is not valid), the cache management unit 142 Is issued to the file input / output response means 134 of the server computer 100.

In response to the file access request, the file input / output response unit 134 of the server computer 100 accesses the file in the disk device 136, reads the corresponding block, and transfers the block to the gateway computer 110.

The cache management unit 142 stores the block in the disk cache 144, and sets or updates the block information for the block in the disk cache 144. At the same time, the cache management unit 142 transfers the block to the file input / output request unit 154 of the client computer 120.

The operation of this system when there is no valid file in the disk cache 144 has been described above.

Next time, when there is an access request for the same file, the cache management unit 142 of the gateway computer 110 fetches the block from the disk cache 144 and sends it to the file input / output request unit 154 of the client computer 120 that issued the access request. To transfer the block. Therefore, in this case, a transfer request from the gateway computer 110 to the server computer 100 is not issued. Therefore, the speed of the relay access is increased.

A similar technique is disclosed in Japanese Patent Application Laid-Open No. 63-200244 (named "File Access System") and Japanese Patent Application Laid-Open No. 63-20184 ("Remote File Access System"). It has been disclosed. In any of the techniques, when an access request for a file object is issued, the modification time of the file object held by the server computer is checked in order to confirm whether the content of the file object has been updated. Compare with the change time. If the contents of the cache file are old, the client computer reads the updated contents of the server computer. However, in the techniques described in these two publications, the cache file exists in the client computer, and the gateway computer is included in the system as in the system disclosed in JP-A-4-313126. Not necessarily.

However, these three prior art documents have a common feature that the content of the file object of the server computer and the content of the cached file object are not conflicted.

防 ぐ Preserving the integrity of files that should be the same in this way is called prevention of discrepancies. The data of the file in which the discrepancy has occurred is called stale data.

The basics of the above-described conventional technology use an algorithm for checking the change time of a corresponding file object of a server computer via a network and comparing the change time with the change time of the contents of a corresponding cache file. Therefore, the server computer must be inquired via the network. If the effective transfer speed of the network is low, or if the load on the server computer having the corresponding file object is too high to respond immediately, such a query itself may take a considerable time. Therefore, some network file systems do not employ such a file cache system. That is, in such a system, the corresponding file object in the cache of the gateway computer is not necessarily updated even though the file object held by the server computer is updated. As a typical example of such a network file system, there is a widely distributed multimedia information system using HTTP (Hyper Text Transfer Protocol) in the so-called Internet.

The Internet is a global network that uses the TCP / IP protocol as its basic protocol. A regional distributed multimedia information providing system built on the Internet is called World Wide Wed (WWW). WWW can handle file objects distributed on a network. These file objects include not only text data but also various types such as image data, audio data, and video image data. Since WWW is attractive to both information providers and information users (users), traffic related to WWW on the network is explosively increasing. In the WWW system, a user of a client computer simply executes browser software having a graphical user interface on the client computer, and successively transmits information composed of various file objects held by a server computer distributed on a network. Can be accessed. WWW has been remarkably popular recently due to such simplicity of operation.

And this WWW system transfers file objects by the HTTP protocol built on the TCP / IP protocol.

In implementing the HTTP protocol, relaying and transferring file objects by a gateway computer as described in JP-A-4-313126 is widely performed. The data is cached on the gateway device. Further, in the WWW system, occurrence of stall data in relay transfer of a file object is permitted. That is, the case where the content cached in the gateway computer does not match the file object of the server computer is permitted.

ス The ratio of the stale data in the cache file is called the stale data ratio. Unless the contents of the cache file are updated in any way, the steal data rate will increase. Generally, when the steal data rate of the cache file increases, the user operating the client computer voluntarily determines that the content is old, and tries to load the file object again from the server computer. In this case, a protocol that does not use the contents of the cache file is used. As a result, the cache file system in the gateway computer tends to be meaningless.

On the other hand, there is an advantage of a protocol that allows the generation of such stale data. If the client computer requests access to the file object of the server computer by relaying it to the gateway computer, and accesses the cache file of the gateway computer, the file object is read from the cache file and transferred to the client computer. Is done. It is not necessary to inquire the server device about the change time of the file object. Therefore, the file object is retrieved from the cache in a short time and returned to the client.

Network file protocols that accept steal data, such as the HTTP protocol, have advantages when used in global wide area networks. That is, if the file object is cached in the cache of the gateway device, no access to the server device occurs. As a result, there is an advantage that the response time can be shortened.

As described above, a gateway computer that speeds up access to file objects on a network and reduces Internet traffic is also particularly called a “proxy server device”.

“Proxy” means “agent”. As the name implies, when a client computer accesses a file object of a server computer on a network, the proxy server device receives the access request on behalf of the server computer and relays the transfer of the file object. The concept of the proxy server device will be described with reference to FIG.

The @proxy server device 13 is provided, for example, between an internal network 23 used in a certain company and an external network (such as the Internet) 25 outside the company. The internal network 23 includes a plurality of client computers 24, and the external network 25 also includes a plurality of server computers 11. Each server computer 11 has a file object 12.

The proxy server device 13 receives the cache file 16 and the read request 19 from the client computer, and if the file object exists in the cache file 16, returns the file object as data 20, and stores the file object in the cache file 16. If there is no valid file object, the proxy process 14 issues a read request 17 to the server computer 11 via the external network 25, receives the corresponding file object as data 18, and returns it to the client computer 24. And an access log 15 for recording the transfer record of the file object by the proxy process 14. The proxy process 14 has a function of newly writing the file object in the cache file 16 when the file object is acquired from the server computer 11.

This proxy server device 13 operates as follows. First, the client computer 24 in the internal network 23 issues a read request 19 for the file object 12 of the server computer 11 on the external network 25 to the proxy process 14 in the proxy server device 13. In response to the read request 19, the proxy process 14 accesses the cache file 16 and determines whether or not the cache file contains cache data of the file object. If the file object is cached, the validity period of the cache file unique to the proxy server device 13 is compared with the last modification time of the cached file object. If the cache has not expired, the file object is read from the cache file (22), and the data 20 is returned to the client computer 24 which issued the read request 19 of the internal network 23.

(4) If the corresponding file object does not exist in the cache file 16, and if the validity period has passed even if it does exist, it is determined that there is no valid file object in the cache file 16. In this case, the proxy process 14 issues a read request 17 for the file object 12 to the server computer 11 on the external network 25. In response, the server computer 11 returns the file object 12 as data 18 to the proxy process 14. The proxy process 14 stores, as an access log in the log file 15, a file object name (that is, a set of a network protocol name, a network address name of a server computer, and a file name), and the substance of the file object data (that is, the file object itself) ) And the writing time are written (26).

The $ proxy process 14 further transfers the data 20 to the client computer 24 in the internal network 23, and writes the file object in the cache file 16 (21).

Therefore, in the cache file 16, the file object name, the entity of the file object, and the time when the file object was obtained from the server device (the last change time) are recorded.

The proxy server device is a computer running UNIX (R) that connects software such as httpd developed by the European Institute for Nuclear Physics CERN, and DeleGate developed by Yutaka Sato of the Electronic Technology Research Institute of Japan. It is generally realized by executing on the above.

FIG. 17 shows the physical configuration of the $ proxy server device. The proxy server device includes a workstation 200 on which UNIX (R) operates. The workstation 200 includes a CPU (Central Processing Unit) 202, a memory 206 connected to the CPU 202 by an internal bus 204, an I / O (input / output) device 208 for files, and an external network via a router 210. And a second network I / O interface 214 connected to the internal network. The I / O device 208 is connected to a file unit 216 used as a storage location for cache files, access logs, and various work files.

The workstation 200 implements the proxy server device 13 by the CPU 202 executing the above-described software such as httpd and DeleteGate.

FIG. 18 is a diagram showing another configuration example of the proxy server device. The proxy server device comprises a workstation 300. The workstation 300 has a configuration substantially similar to that of the workstation 200 in FIG. 17, but includes a serial port 302 connected to a modem device 304 instead of the second network I / O interface 214 in FIG. 17. Is different. The workstation 300 is connected to an internal network via a modem device 304 and a public telephone network 306.

に お い て In FIG. 18, the same parts as those in FIG. 17 are denoted by the same reference numerals. Their names and functions are also the same. Therefore, detailed description thereof will not be repeated here.

FIG. 19 shows still another configuration example of the proxy server device. The proxy server device of FIG. 19 is also realized by the workstation 400. The workstation 400 has a configuration similar to that of the workstation 200 shown in FIG. 17 except that the workstation 400 is connected to the internal network 23 instead of the first network I / O interface 212 and the second network I / O interface 214. In that it has a single network I / O interface 402. The internal network 23 is connected to an external network 25 via a router 210, and is also connected to a plurality of client computers 24. The workstation 400 can communicate with the server computer 11 on the external network 25 via the internal network 23 and the router 210, and can communicate with the user 31 via the internal network 23 and the client computer 24.

と In FIGS. 19 and 17, the same components are given the same reference numerals and names. Their functions are the same. Therefore, detailed description thereof will not be repeated here.

The @proxy server device has a relay data caching mechanism as described above. That is, the proxy server device caches the data of the file object to be relayed in the cache file device existing therein.

The cache server of the proxy server device has an expiration date unique to the proxy server device. If a request to access a file object on the same server computer is received from a client computer within the expiration date after the file object has been written to the cache, the file object previously written to the cache file is retrieved. Transfer to client computer. Thus, access to the server computer does not occur.

(4) The effect of the cache by such a gateway device will be specifically described below. The proxy server device is installed in a certain company organization, and connects to an external network at a speed of 64 kbps. In addition, this proxy server device is connected to a local area network in a company organization at a speed of 10 Mbps. The speed difference between the external and internal networks is more than an order of magnitude. Such an example is a common example, and a typical example is the configuration shown in FIG.

At this time, assuming that the relay file object is cached in the proxy server device, the hit rate of the cache is set to Hitrate. Hitrate is an exponent that is 1 when the cache hits 100% for an access request for a file object and becomes 0 when the cache hits 0%.

Note that the speed of reading data from the cache file and writing data to the cache file in the proxy server device is sufficiently higher than the network speed. That is, it is assumed that the time required for inputting / outputting the file object to / from the cache file can be ignored.

{At this time, the average access speed Vave for accessing the file object in the external network from the inside is represented by the following equation 1.

If no cache file exists, Hitrate is 0. Therefore, the average speed (kbps) is equal to the access speed of the external network of 64 kbps from Equation 1. If Hitrate is 1, that is, if the cache file hits 100%, the average speed (kbps) is 10,000 kbps (= 10 Mbps) equal to the speed of the internal network. In other words, the higher the hit rate, the higher the average access speed seen from the client computer. FIG. 20 is a graph of Equation 1 showing the relationship between the hit ratio and the average access speed.

By the way, the cache hit rate depends on the amount of the cache file. That is, if as many file objects as possible are stored in the cache file, the probability of being included in the accessed file object or cache file again is high, and the hit rate is high. FIG. 21 is a graph showing the relationship between the cache hit ratio and the amount of stored cache files as an example of such a relationship. As shown in FIG. 21, when the cache file size is reduced, the cache hit rate is close to 0, and when the cache file size is increased, the cache hit rate gradually increases.

In the $ proxy process, as a method of maintaining the contents of the cache file, a control method of invalidating a file object that has been passed for a predetermined period since writing by the first file access occurs. This fixed period is called an expiration date. The person who manages the proxy server device sets an appropriate value for this expiration date according to the actual situation. If the expiration date is set longer, the amount of data stored in the cache file increases. As a result, the cache hit rate increases. That is, to increase the cache hit rate, the cache expiration date may be extended.

フ ァ イ ル Since the storage capacity of the file device for storing the cache file is limited, the expiration date cannot be extended without limit. An expired cache file is deleted at some timing. In this specification, the time until the cache file is erased is referred to as an erase time.

According to the experiment performed by the inventor of the present application, the cache expiration date is substantially proportional to the amount of files stored in the cache file. This relationship is shown in FIG. This is because the amount of access from the internal network to the external network is almost constant regardless of the day. Therefore, the following relationship is established.

関係 From this relationship, as shown in FIG. 22, it can be seen that increasing the cache expiration date increases the cache hit rate.

(4) The relationship between the cache expiration date and the amount of file objects differs depending on the frequency with which the internal users using the client computer access the file objects of the server computer on the external network. Although the cache expiration date and the cache file storage amount are almost proportional, the proportional coefficient varies depending on the use environment such as the network speed and the number of internal client computer users.

(4) The cache hit rate is about 12% when the validity period of the cache file is about one day. Experience has shown that a cache hit rate of about 41% with an expiration date of 14 days. When the expiration date is 14 days, it has been observed that about 700 MB of the file object is stored in the cache file.

In addition, as the size of the local network increases in recent years, and the number of connected client computers increases, or as the read requests of the client computers increase, a single proxy server computer is not sufficient for the requests from each client computer. It's getting harder to respond. Therefore, a method of improving the efficiency of access processing to a file object by combining a plurality of proxy server computers has been proposed and implemented.

FIG. 23 is a configuration example of a system using a plurality of proxy server computers. The proxy server computer 501 relays between the wide area network and the local network, and caches file objects. This proxy server computer 501 is hereinafter referred to as a secondary cache proxy server computer. On the other hand, the proxy server computers 503 to 508 are called primary cache proxy server computers, and relay and file objects between the client computer 24 and the secondary cache proxy server computer 501 connected to the divided local network. Cache.

For example, assume that a read request for a file object is output from the client computer 24 to the primary cache proxy server computer 504. At this time, if a valid file object exists in the cache file 510 of the primary cache proxy server computer 504, the primary cache proxy server computer 504 fetches the file object from the cache file 510 and issues a read request to the client computer. 24.

If the valid file object does not exist in the primary cache proxy server computer 504, the read request from the client computer 24 is relayed to the secondary cache proxy server computer 501. The secondary cache proxy server computer 501 similarly checks the presence or absence of a valid file object in the cache file 502, and sends the file object to the primary cache proxy server computer 504 if the file object exists. Through the wide area network 61 to the server computer 11.

(4) The subsequent processing is the same as that of a single proxy server computer, and therefore, detailed description thereof will not be repeated here.

According to this method, each of the primary cache proxy server computers 503 to 508 can share and take charge of a client computer, thereby distributing the load. In addition, the secondary cache proxy server computer 501 only needs to perform processing on file objects that have not been cached in each of the primary cache proxy server computers 503 to 508, so that the load can be reduced. Specifically, when there are N client computers 24 in total, the proxy server computer needs to process the requests of the N client computers in the conventional method. It is sufficient to process the requests from the client computers for the number of client computers.

However, in this method, one secondary cache proxy server computer 501 performs all relays with the server computer 11 connected to the wide area network 61, and it cannot be said that the essential load distribution is achieved. .

There is also a lot of waste in using cache files. That is, it is considered that all the contents of the file objects of the primary cache proxy server computer groups 503 to 508 are also held by the secondary cache proxy server computer 501. This is because the file objects are always cached in the secondary cache proxy server computer 501 and the primary cache proxy server computers 503 to 508 on the relay route. In addition, even if each primary cache proxy server computer reads the same file object from a client computer connected to another primary cache proxy server computer, the same file object may be duplicated. Is high.

Therefore, it can be said that the cache capacity of the entire distributed file system is the capacity of the cache file 502 of the secondary cache proxy server computer 501, and there is a limit to the cache file capacity with only one secondary cache proxy server computer. It can be said that.

方式 In order to solve such a problem, a system as shown in FIG. 24 has been proposed and implemented.

In this method, there is no proxy server computer corresponding to the secondary cache proxy server computer 501 in FIG. Each of the proxy server computers 601 to 606 is in charge of a group of client computers. For example, a read request from the client computer 24 connected to the proxy server computer 602 is received by the proxy server computer 602. Here, when a valid file object exists in the cache file 608 of the proxy server computer 602, the file object is extracted and sent to the client computer 24 that issued the read request.

If there is no valid file object in the cache file 608 of the proxy server computer 602, the proxy server computer 602 is connected to each of the other proxy server computers 601, 603 to 606 connected to the local network 600. It is inquired whether a valid file object is held in the cache files 607, 609 to 612. When there is a proxy server computer holding a valid file object, for example, when the proxy server computer 605 holds a valid file object in the cache file 611, the proxy server computer 605 becomes a proxy server computer 602. , The proxy server computer 602 sends the file object to the client computer 24 that issued the read request.

If there is no valid file object in the cache files 607 and 609 to 612 of the proxy server computers 601 and 603 to 606 connected to the local network 600, the proxy server computer 602 transmits the server computer via the wide area network 61. 11, the read request from the client computer 24 is relayed.

(4) The subsequent processing is the same as that of a single proxy server computer, and thus detailed description thereof will not be repeated.

(4) According to this method, the same file object is not held by a plurality of proxy server computers, and the effective use of the cache file can be achieved. Further, the cache capacity of the entire distributed file system is the sum of the capacities of the cache files 607 to 612 of all the proxy server computers 601 to 606, so that the cache capacity can be increased. Therefore, a high cache hit rate can be expected, and an improvement in the execution transfer speed can be expected.

方式 Also, a method as shown in FIG. 25 has been proposed and implemented. 23 is different from the method of FIG. 23 in that a plurality of secondary cache proxy server computers 705 to 707 are prepared, and primary cache proxy server computers 711 to 716 are read from the client computer group by the secondary cache proxy server computers 705 to 707. When the request is relayed, for example, if the name is xxx.xxx depending on the address name of the server computer 701 or 702 to which the read request is requested. If it is edu, it judges that the request is a read request to the edu domain server computer 702, and outputs an access request to the proxy server computer 705. The name is yyy. If it is “com”, the access request is determined to be a read request to the com domain server computer 702, and an access request is output to the proxy server computer 706. As described above, the secondary cache proxy server computers 705 to 707 that relay the access request by a simple rule based on the end domain name (edu, com, etc.) are selected.

{Other processing is the same as that of a single proxy server computer, and therefore, detailed description thereof will not be repeated.

As a result, read requests do not concentrate on the same secondary cache proxy server computer, and the load can be distributed. Also in this method, it is not necessary for a plurality of secondary cache proxy server computers to hold the same file object, and the cache file can be effectively used. Further, the cache capacity of the entire distributed file system is the sum of the cache file capacities of all the secondary cache proxy server computers, and the cache capacity can be increased. Therefore, a high cache hit rate can be expected, and an improvement in the execution transfer speed can be expected.
JP-A-4-313126 JP-A-63-200244 JP-A-63-20184

In the above proxy server device, by setting the cache expiration date to a certain length, the capacity of file objects stored in the cache file increases, the cache hit rate increases, and efficient file object transfer becomes possible. .

However, on the other hand, the contents of the file object in the server computer may be modified and changed. Therefore, if the expiration date of the cache file is long, the contents of the file object in the server computer may be changed during that period. Regardless, the user of the client computer in the internal network reads out the file object having the old contents stored in the cache file.

For example, many file objects such as weather charts and electronic newspapers have the same file object name but are updated every few hours. Therefore, for example, if the cache expiration date is set to 24 hours, information such as weather charts and electronic newspapers updated every three hours becomes old and useless information with a probability of 7/8 or more.

FIG. 26 is a time chart showing the data flow of cache control in such a conventional proxy server device. Names such as A, B, C, D, and E indicate that they are data having the same file object name but different contents. When the client computer accesses (user access (1)) via the proxy server device, the content B of the file object of the server computer is cached by the proxy server computer, and B is transferred to the client computer. Next, when the user access (2) is transmitted from the client computer, since the content B cached in the proxy server device is valid for 24 hours, a cache hit occurs and the content B of the file object is transferred to the client computer. However, at this time, although the content is already E in the server computer, the client computer sees old data having the content B.

(5) Among the contents of such a cache file, the rate at which the original data existing in the server computer connected to the external network is changed is called the stale date rate. Therefore, increasing the cache expiration date increases the capacity of file objects stored in the cache file and increases the cache hit rate, but increases the stale data rate of the contents in the cache file.

With a proxy server device composed of the above-mentioned CERN httpd software, the cache expiration date of a file object of a specific server computer can be reduced. However, in this method, when a client computer accesses a file object, the cache expiration date is reduced. Is short, the server computer accesses the file object after a mishit in the cache. Therefore, the user of the client computer cannot receive the benefit of the cache by the proxy server computer and waits until the low-speed file object transfer by the wide area network is completed.

Therefore, in the above-described conventional technology, the freshness of the content of the file object in the cache file and the maintenance of the cache hit ratio are mutually contradictory problems.

Generally, when the steal data rate of the cache file increases, the user of the client computer voluntarily determines that the contents should be old and loads the file object directly from the server computer. In this case, a protocol is used that specifies that the data should be actually loaded from the server computer, passing through the cache file. As a result, the cache file of the proxy server computer becomes meaningless.

As a method of distributing the load on the proxy server computer, the methods shown in FIGS. 23 to 25 have been proposed as described above. However, in any case, it cannot be said that the load is sufficiently distributed. In the method shown in FIG. 23, since one secondary cache proxy server computer 501 performs all relays with server computers connected to the wide area network, it cannot be said that load distribution is achieved.

In the method of FIG. 24, the paths between the proxy server computers 601 to 606 and the wide area network 61 are not multiplexed. Also, the proxy server computers 601 to 606 connected to the local network 600 need to be replaced with those that understand a special protocol corresponding to this method.

In the method of FIG. 25, the primary cache proxy server computers 711 to 716 perform distribution according to simple rules such as domain names, so that the load is uneven among the secondary cache proxy server computers 705 to 707.

SUMMARY OF THE INVENTION The present invention has been made to solve the above problems, and an object of the present invention is to distribute a request for accessing a file object, effectively use a path, improve a throughput, and further reduce a traffic on a local network. It is to reduce.

According to the first aspect of the present invention, a client computer, a server computer that provides information in response to an information provision request from the client computer, a gateway device that controls a file object of the server computer, and an information provision request from the client computer Device in a system consisting of a proxy server computer that relays the transfer of file objects based on the information, and a calculation result by a predetermined calculation using a character string included in the information provision request and the number of the proxy server computers in the system , A proxy server computer that stores the requested information is determined, thereby controlling the amount of information stored for each proxy server computer.

The invention according to claim 2 is the gateway device according to claim 1, wherein the predetermined operation is performed by adding the character code of the URL character string included in the information provision request and calculating the number of proxy server computers in the system. It is a hash operation using the remainder of the division.

The invention according to claim 3 is the gateway device according to claim 1 or 2, wherein the predetermined calculation uses the same formula as another gateway device in the system, and the calculation result by the same information provision request is It is characterized in that the same proxy server computer is determined.

According to a fourth aspect of the present invention, there is provided a client computer, a server computer for providing information in response to an information provision request from the client computer, a file object control of the server computer, and a file object based on the information provision request from the client computer. A proxy server computer in a system comprising a proxy server computer for relaying the transfer of the information, and a calculation result obtained by a predetermined calculation using a character string included in the information provision request and the number of the proxy server computers in the system. Based on this, the proxy server computers that store the requested information are determined by including the own device, thereby controlling the amount of information stored for each proxy server computer to be dispersed.

The invention according to claim 5 is the proxy server computer according to claim 4, wherein the predetermined operation is to add the character code of the URL character string included in the information provision request and to determine the number of proxy server computers in the system. Is a hash operation using the remainder divided by.

According to a sixth aspect of the present invention, there is provided the proxy server computer according to the fourth or fifth aspect, wherein the predetermined operation uses the same calculation formula as another proxy server computer in the system and is operated by the same information provision request. The result is characterized by determining the same proxy server computer.

According to a seventh aspect of the present invention, there is provided a client computer, a server computer for providing information in response to an information provision request from the client computer, and a proxy server computer for relaying a transfer of a file object based on the information provision request from the client computer. A proxy server that stores requested information based on a calculation result of a predetermined calculation using a character string included in the information provision request and the number of proxy server computers in the system. By determining the computer, the amount of information stored for each proxy server computer is controlled to be dispersed.

The invention according to claim 8 is the client computer according to claim 7, wherein the predetermined operation is to add a character code of a URL character string included in the information provision request and calculate the number of proxy server computers in the system. It is a hash operation using the remainder of the division.

According to a ninth aspect of the present invention, there is provided the client computer according to the seventh or eighth aspect, wherein the predetermined operation uses the same calculation formula as another client computer in the system, and the operation result by the same information provision request is It is characterized in that the same proxy server computer is determined.

[Embodiment 1]
A gateway device (referred to as a proxy server device) to which the present invention is applied can be easily realized by a computer running a UNIX® OS having a network interface, with reference to the following description and drawings. An example of the configuration of such a proxy server device is as shown in FIG. Since the proxy server device of FIG. 19 has already been described, detailed description of its configuration will not be repeated here.

Normally, the user 31 sends a request via the network 23 to the proxy server device (the proxy server device 400 in FIG. 19) using the client computer 24 to acquire the file object 10 of the server computer 11. . As described above, the file object 12 includes a set of the protocol name, the network address of the server computer, the name of the file object, and the file data entity.

FIG. 1 shows a system configuration of the device 53 in the first embodiment of the proxy server device according to the present invention. The proxy server device 53 includes a first proxy process 14 connected to the internal network 23 and the external network 25, a cache file 16 used by the first proxy process 14, and an access log ( An access log 15 for recording a file name as a transfer record), a FIFO buffer 35 for sequentially taking out and storing the access log 15 from the end thereof, and sequentially outputting the stored access log, and an access pattern list created in advance Is stored in the memory buffer URLBUF (URL will be described later) if the access log is extracted from the FIFO buffer 35 and compared with the access pattern stored in the pattern file 30 and they match. And a second proxy process 34 that acquires the file object from the server computer 11 and stores the file object in the cache file 16 based on the file object name included in the access log stored in the URL BUF. Including.

Each of the first proxy process 14, the second proxy process 34, and the cache update process 36 is stored as a program in the memory 206 shown in FIG. 19, for example, and is realized by the CPU 202 executing the program. .

The first proxy process 14 corresponds to a transfer unit described in claims, the cache file 16 corresponds to a cache file unit, and the access log 15 corresponds to a file transfer recording unit. The cache update unit 36 includes the cache update process 36, the pattern file 30, the FIFO buffer 35, and the second proxy process 34.

The second proxy process 34 is specific to the gateway device according to the present invention, and is activated under the following conditions.

{(1)} The second proxy process 34 uses the same cache file 16 as the first proxy process 14.

{(2)} The cache expiration date of the second proxy process 34 is set to 0. That is, access to the server computer 11 via the second proxy process is always performed to the server computer 11 in the external network 25, and a file object is acquired and stored in the cache file 16.

The operation of the proxy server device 53 shown in FIG. 1 will be described below. It is assumed that the client computer 24 in the internal network 23 sends a network access request 19 to the proxy server device 53. The first proxy process 14 receives the network access request 19, and first accesses the cache file 16 existing in the file device 216 via the I / O interface 202 (see FIGS. 17, 18, and 19). (21). Further, the first proxy process 14 compares the network access request with the file object name in the cache file 16 (22), and determines whether or not they match. If there is a match, the first proxy process 14 further extracts the last modification time of the file object in the cache file 16 (22) and compares it with the current time. If the current time is within the expiration date starting from the last change time of the file object, the file object is extracted from the cache file 16 (22), and is sent to the client computer via the network I / O interface 402. The file object is returned to 24 (20).

If the file object does not exist in the cache file 16, or if the cache object has expired, the first proxy process 14 sends the file object via the network I / O interface 402. Then, a file object transfer request 17 is sent to the server computer 11 on the external network 25. At this time, the server computer 11 is specified by the network address name in the file object.

(4) The corresponding server computer 11 returns the requested file object to the first proxy process 14 (18). That is, the file object is acquired by the first proxy process 14.

The first proxy process 14 writes this file object in the cache file 16 together with its last modification time (21), and records the name of this file object in the access log 15 (26).

On the other hand, the proxy server device 53 of the present invention is different from the conventional proxy server device in that the cache update process 36 and the second proxy process 34 are processed and run by the CPU 202 to update file objects.

The cache update process 36 and the second proxy process 34 operate as follows. The access logs stored in the access log file 15 by the first proxy process 14 are sequentially stored in the FIFO buffer 35 from the end (28). The FIFO buffer 35 sequentially outputs the access log stored first (27). The cache update process 36 extracts the access log line by line from the FIFO buffer 35 and stores it in the memory buffer. The cache update process 36 extracts the access log from the memory buffer and determines whether or not the cache has been hit based on each access log. If the cache hits, an access pattern list is further extracted from the pattern file 30 (29), and a file whose file object name included in the access log matches the access pattern list is extracted. In addition, the cache update process 36 excludes the extracted file objects whose file object names match the specific pattern, and stores them in the memory buffer URLBUF.

If the URL BUFF is not empty, the second proxy process 34 accesses the server computer 11 (32) based on the file object name stored in the URL BUF (32) and acquires the file object (33). ) Store in the cache file 16 (37). As a result, the file object in the cache file 16 is maintained at the latest content.

起動 Thus, when the second proxy process 34 is activated, the following two effects can be largely obtained.

(1) In the first proxy process 14, the expiration date of the cache file is set to a certain value. If the cache update processing of the file object is performed by using the first proxy process 14, the cache file 16 There is a high possibility of hitting. In this case, since the access to the file object of the server computer 11 on the external network 25 is not performed, the update of the cache file 16 cannot be realized.

On the other hand, the second proxy process 36 has a cache expiration date of 0, and if the second proxy process 36 issues an access request to the second proxy process 36, this cache file is a mishit. (32), the file object is transferred to the second proxy process 34 (33), and the file object in the cache file 16 is updated to the latest state. Be updated.

{(2)} The first proxy process 14 leaves an access log, and a file object to be accessed for cache update is extracted based on the access log. Therefore, if the cache update process 36 accesses the file object of the server computer 11 using the first proxy process 14, an access log generated as a result is recorded in the access file 15. The cache update process 36 extracts a file object name to be accessed based on the access log stored in the access log file 15 in the next update access operation. Therefore, the same file object is obtained again from the server computer 11, and a loop operation in which the same file object appears multiple times occurs. In order to avoid this, the second proxy process 34 is operated, and a cache update access operation using the second proxy process 34 is performed.

A WWW system on the Internet is an example to which the present invention can be applied very effectively. The procedure for performing the cache update process of the present invention in the proxy server device on the WWW system will be described below.

In the WWW system, file object names distributed on a network are expressed and specified in a format called Uniform Resource Locator (URL). An example of the URL is shown below.

に お い て In Equation 4, “http” indicates a protocol to be used. “Www.xxx.co.jp” indicates the address of the HTTP server computer on the network, and is selected so that there is no identical one on the network. “/Test/index.html” indicates a file name in the server computer.

When using generally used DeleGate as the first proxy process 14, examples of the access log include the client computer name, time, "HTTP protocol (file acquisition request)", other information, and cache. Hit status (H if cache hit). The following is an example of the access log.

The access log shown in Equation 5 corresponds to an access log when the applicant accesses an information page published on the Internet as shown in FIG. Since the page shown in FIG. 2 is composed of text and six pieces of graphic data, once access to this page is made, seven access logs including access to six pieces of graphic data are formed. You. That is, in the above access log, the first line is an access request to the text of this information page, and the remaining six lines are graphic data (sharpcolor.gif) of the logo of the applicant company pasted on the cover and the applicant. Acquisition of a logo (isg.gif) of a division called Information System Business Division in the company, and graphic data of photographs of four products of the applicant company (Zarus.gif, Shoin.gif, Prostation.gif, S2.gif) Indicates a request.

The HTTP protocol is composed of a command character string (such as GET), a URL, and a protocol version (such as "HTTP / 1.0"). The following is an example.

The HTTP protocol is described in References 3 and 4 among the references shown in the last table. The procedure for performing the cache update process will be described below. The prerequisites for that are as follows (1) and (2).

{(1)} The first proxy process 14 starts with a cache expiration time M (time). For example, M = 24 hours.

{(2)} The second proxy process 34 is activated according to two conditions: (i) sharing the cache file with that of the first proxy process 14 and (ii) setting the cache expiration date to 0.

(5) Under such conditions, the cache update process of the present invention is started at a fixed time every day and ends at a fixed time. To automatically start the cache update process at a fixed time every day, a timer function (cron) provided in UNIX® OS is used. This is a well-known function related to the UNIX (R) OS, and therefore, its details will not be described here. It should be noted that similar functions are often provided in OSs other than UNIX (R) @OS.

FIG. 3 is a flowchart showing the processing procedure of the cache update process of the present invention. A program for implementing this process is arranged, for example, in the memory 206 of FIG. A network I / O interface 402 and an I / O interface 208 are used for input and output between this process and the network and file device 216, respectively.

First, in step S1, the first proxy process 14 reads one line from the end of the access log stored in the access log file 15. If it cannot be read, it waits until the access log can be read. In UNIX (R), such a function of calling the end of the access log can be performed by a command tail provided with the OS, and is a character string buffer secured in a memory by reading one line from its standard output. To store one line in the LINE buffer, the following is performed. Here, the access log stored by the first proxy process 14 is / usr / local / etc / telegate / logs / 10000. http. This corresponds to the access log stored in the access log file 15 of FIG.

1-row reading by {7 tail} -f corresponds to 27 in FIG. 1. This constitutes a FIFO buffer 35 by software, and has a read pointer and a write pointer, respectively.

{1} That is, the access log written by the first proxy process 14 occurs one line for each transfer of one file object. At this time, the write pointer (wp) is advanced by one line. If the FIFO buffer 35 by software reads one line of the access log by tail $ -f, the read pointer (rp) is advanced by one line. The writing speed (line / second) by the first proxy process 14 and the one-line reading speed (line / second) by tail # -f in the cache update process 36 may not be the same. When the access log writing speed by the first proxy process 14 increases, the write pointer (wp) advances, but the access log is read by the cache update process 34 while the read pointer (rp) follows it. If the access log writing speed is low or if the processing of the cache update process 34 on the reading side progresses, the read pointer (rp) will eventually catch up with the write pointer (wp). This is the operation as a software FIFO buffer.

Next, since the one-line access log information stored in the LINE buffer is a record of a file object that has been accessed in the past, a file containing a file object name to be accessed for cache update under the following conditions is selected. . This selection process is hereinafter referred to as filtering, and corresponds to step S2 in the flowchart shown in FIG. The rules for this filtering are as follows.

{(1)} An HTTP command character string is a read request (GET) for a file object.

{(2)} The protocol part of the file object name is http.

{(3)} An access that hits the cache.

{(4)} Exclude files whose file object names are not suitable for cache update access.

{(5)} An access list including a specific character string pattern matches a file object name.

(4) Filtering is performed using the above conditions (1) to (5), only the file object name field is extracted, and the file object name is extracted by cache update access.

Only the protocol part of the file object name which is http is extracted because only the http protocol is used by the DeleteGate software generally used as the first proxy process 14 and the second proxy process 34 to generate a cache file. This is because only that file object is meaningful for cache update access. Since access by other protocols is not cached, there is no point in performing cache update access.

The reason that only those that hit the cache are extracted is that it is interpreted that the file object that has not hit the cache has been accessed to the server computer 11, so it is not necessary to perform the cache update access immediately after that. It is because it is judged.

File objects whose names are not suitable for cache update access include those that include “?” In the URL. Since the URL notation including “?” Is used when a user using the client computer 24 manually inputs a character string to the server computer 11, the file object having the URL notation including this character is updated in the cache. If accessed, unexpected results may occur.

Only those that match the access list that contains the specific character string pattern are selected by the server computer that provides the weather map, electronic newspaper, etc., whose contents are frequently updated in the server computer. There are many. Therefore, by limiting only those that match the access list that stores such a character string pattern that specifies the file object of the server computer to the cache update access, unnecessary cache update access is reduced as much as possible. Things.

{The pattern file 30 storing such a character string pattern is stored in the pattern. txt, and its contents are shown in Expression 9.

In UNIX (R), such a file object name extraction operation can be performed using Expression 10 which is a command group provided with the OS.

{For example, by inputting the following command, the one that has undergone the processing in step S2 is output to the character string buffer URLBUF on the memory.

If the above filtering rule is violated, the character string buffer URLBUF becomes empty. The progress during this time will be described below. The format of the access log file 15 created by the first proxy process 14 is as follows, as described above.

@ When a cache hit occurs, the character "H" is output at the end of the line as a cache hit status.

First, by inputting grep -v “＼?”,? Extract the lines that do not contain and output to the standard output. Further, only the line which is a GET command, which is a file object acquisition request, is extracted with grep "@" GET "and output to the standard output. , That is, select a hit in the cache and output it to the standard output. Here, ＄ is a regular expression representing the end of a line.

Assuming that the blank is a delimiter between the fields, the seventh field corresponds to the file object name, so that only the file object name is extracted by using awk 'print' 7 '.

{Furthermore, the file object name part is a pattern file pattern. txt and whether or not it matches any of the lines in fgrep {-f} pattern. The matching is performed by txt and the result is output to the character string buffer URLBUF.

In step S3, it is determined whether or not the character string buffer URLBUF is not empty. If the character string buffer URLBUF is empty, the process returns to step S1, and if not, the process proceeds to step S4.

In step S4, the second proxy process 34 acquires the file object extracted in the cache update process 36 from the server computer 11. The second proxy process 34 activates a child process for accessing the server computer 11 based on the file object name for cache update access stored in the character string buffer URLBUF. The child process is called a cache update child process. The cache updater process makes a network connection to the server computer 11 through the second proxy process 34. As a result, access to the file object stored in the server computer 11 occurs, and the content of the cache file 16 is updated. Details of this processing will be described later.

In step S5, the current time is checked, and if the specified time has passed, the process ends. If the specified time has not passed, the process returns to step S1, and the above processing is repeated.

The above is the algorithm in the proxy server device according to the first embodiment of the present invention. FIG. 7 shows that the data of the file object in the case where the cache expiration date of the proxy server computer 53 is 24 hours and the contents are updated every three hours while the file object of the server computer 11 has the same name, is cached. Is shown.

First, if there is no corresponding file object in the cache file 16 when the user access (1) occurs, the data B is accessed from the server computer 11 and written into the cache file 16 and transferred to the client computer 24. Is done.

Next, when the user access (2) occurs, the data B hit in the cache is transferred to the client computer 24. At this time, since the server computer 11 has the data D, the client computer 24 has read the old stale data of about 3 hours. However, as a result of this operation, the access log including the file object name and the information that the cache hit has been recorded in the access log file 15, and the proxy server device 53 voluntarily operates the server according to the operation described with reference to FIG. Access the device 11, read the data E, and update the cache file 16.

(4) Further, when the user access (3) occurs, the cached data E is transferred to the client computer 24. This operation triggers the access log including the file object name and the information that the cache hit has been recorded in the access log file 15, and the proxy server device 53 voluntarily operates the server according to the operation described with reference to FIG. It accesses the computer 11, reads out the data I, and updates the cache file 16.

Therefore, when the user access (4) occurs, the cached data I is transferred to the client computer 24.

As can be seen from the above operation, the client computer 24 is still reading the stale data, but it is reading information that is closer to the latest than the case of FIG. If the interval between accesses by the user is within 3 hours, the probability of occurrence of stale data is extremely low, and it can be understood that the latest information of the server computer 11 can be obtained.

As a method of realizing the network access unit in step S4 in FIG. 3, the HTTP protocol may be directly generated from the cache update process, but a UNIX (R) WWW client program such as lynx widely used is generally used. If used, a cache update process operation for performing network access according to HTTP can be realized without newly manufacturing a cache update process that generates HTTP. The method will be described below.

Since \ lynx is a program that runs on the UNIX (R) OS, the description of UNIX (R) will be used in the following description.

By specifying the following command in $ lynx, the specified URL can be read and written to a file named temp_file.

{Furthermore, access via the proxy process can be used from lynx by setting the environment variable http_proxy of the UNIX® OS. In addition, if the proxy process is proxy software such as DeleGate, access via proxy is possible by specifying proxy in the URL notation. See reference 12 for this.

場合 The case where the latter method is adopted will be described below. If the proxy process is a process using the TCP / IP port number 10001 of the gateway computer whose network address is proxyserver, the file object of the server computer 11 via this proxy process http: // www. xxx. co. jp / test / index. In order to access html, the following command of Expression 14 may be specified.

Therefore, when implementing the present invention, the proxy process may be set to the port number of the second proxy process 34 in the same manner.

The object of the present invention is to keep the cache file 16 managed by the first proxy process 14 up to date by such a cache object update access of the file object (URL). Therefore, temp_file itself is not used and may be discarded. If the OS is UNIX (R), / dev / null is prepared as an empty file, and writing to this file is efficient because no file is actually generated.

Next, a cache management method of the first proxy process 14 and the second proxy process 34 will be described. The cache management method of the proxy process is based on a well-known technique such as DeleGate having a cache control function as described in Reference Documents 5 and 12. The cache management method in DeleteGate is disclosed in the form of source code, and the method will be described below. The present invention can be applied to the first proxy process 14 and the second proxy process 34 as long as the proxy server software with a cache function uses the same cache management method.

As described above, it is assumed that the second proxy process 34 is operating in the system named proxyserver, and accepts a URL acquisition request with TCP / IP number 10001. At this time, when the specified URL notation is stored in the character string buffer URLBUF, the cache update process 36 has described that the file object of the specified URL is stored in the temp_file by executing the command described in Expression 16. .

The communication between the second proxy process 34 and the cache update process 36 is performed by TCP / IP, for example, UNIX (R) message communication.

{For example, in the character string buffer URLBUF, http: // www. sharp. co. jp / sample / test. It is assumed that a character string “html” is stored. When the cache update process 36 executes the command of Expression 16, the cache update process 36 outputs a URL acquisition request to the second proxy process 34. The second proxy process 34 generates a child process, leaves the process of acquiring a URL to the child process, and waits for a URL output request from the cache update process 36 again. The child process generated by the second proxy process 34 for executing the URL acquisition by executing the equation 16 by the cache update process 34 performs the processing shown in the flowchart of FIG.

Hereinafter, the operation of the child process generated by the second proxy process 34 will be described with reference to the flowchart of FIG.

The $ proxy process has a cache file 16, which is created as part of the regular UNIX file system. Assuming that a partition named / cache is used as the cache directory, the protocol name is http and the server computer name www. sharp. co. jp, the actual part of the file object is sample / test. html.

At this time, the cache file attempts to generate a cache file according to the naming rule shown in the following Expression 18.

{Therefore, http: // www. sharp. co. jp / sample / test. The html is a file shown in Expression 19 (S11).

The $ proxy process first converts the file name into the file name shown in Expression 19, and checks whether the file exists by checking whether an existing file can be opened by the OS open ($) system call (S12). If the file does not exist, a cache file having the file name shown in Expression 19 is created in the new file creation mode (S13). In addition, exclusive control is performed on the cache file, and exclusive locking is performed so that other processes cannot write (S14).

Next, a file object specified by the URL is obtained from the server computer 11 via the network (S15). The obtained file object is transferred to the client computer 24, and the file object is written in the cache file having the file name shown in Expression 19 (S16). Subsequently, the cache file is closed to release the exclusive lock (S17).

In the above processing, the cache file is exclusively locked, but this exclusive control can be exclusively locked by using the system call "flock (@) function" when the OS is UNIX (R). The reason for using exclusive control is that the proxy process activates a child process each time one file object is transferred and causes the child process to perform cache control, so that conflict occurs between the child processes for the same cache file system. This is because there are cases.

When writing to the cache file, the file change time Tm is recorded as the attribute of the file (Equation 19). The file update time Tm is always added as an attribute of the file by the OS managing the file system of the UNIX® OS. That is, the cache file uses the file system of the normal OS as it is, and is not a special file system at all.

Next, a case where a cache file already exists will be described (S19 to S26).

{Cache file / cache / http / www. sharp. co. jp / sample / test. The last change time Tm of the html is read from the OS. In the case of UNIX (R), the last change time is obtained by the fstat () system call, and is a function that monotonically increases from the past to the future (S19).

(4) The current time is read from the system call and stored in a variable called Tnow (S20). The current time Tnow is obtained from a system call of the OS. The last change time Tm is compared with the current time Tnow to determine whether or not it is within the cache expiration date (S21). If it is within the cache expiration date, the cache file is shared-exclusive locked and cannot be written by another process, but can be read (S22). Then, the existing cache file is read and transferred to the client computer 24 (S23). Further, the cache file is closed and the shared exclusive lock is released (S17).

If the cache expiration date has passed, exclusive control is performed on the cache file, and exclusive locking is performed so that other processes cannot write (S24). Next, a file object specified by the URL is acquired from the server computer 11 via the network (S25). The obtained file object is transferred to the client computer 24, and the file object is written in the cache file (Equation 19) (S26). Then, the cache file is closed to release the exclusive lock (S17).

The above is the cache control procedure of the well-known proxy process whose source code is generally released. In the present invention, the cache file 16 is stored in the first proxy process 14 and the second proxy process by utilizing this property. Shared by the process 34. Since the second proxy process 34 has a cache expiration date of 0, access to the server computer 11 via the second proxy process 34 can always update the cache file 16 in S4 of the flowchart of FIG. Understand.

In the first embodiment, the operation is started at 7:00 when the use of the client computer 24 starts, and is ended at 21:00 at night when the use of the client computer 24 is reduced. As a result, the cache update process 36 is eliminated during a time period when the client computer 24 is not used, so that the memory resources of the computer employing this control method are not wasted.

As described above, the proxy server computer 53 obtains the file object from the server computer 11 and updates the cache file 16 immediately after the cache hit of the access request for the file object made by the user. When the user accesses the same file object, the latest file object is already stored in the cache file 16. Therefore, even if the cache expiration date is set longer, the probability that the file object of the server computer, which is changed at intervals shorter than the cache expiration date, is reflected in the cache increases.

同一 Also, if the same user accesses the same file object again after a while, the contents of the cache will be updated to the latest one. Also, since only file objects that match a specific pattern are subject to update access, unnecessary cache update access can be reduced by registering server computers with frequently updated file objects in advance, thus suppressing unnecessary traffic. The line can be used effectively.

[Embodiment 2]
Next, a second embodiment of the gateway device of the present invention will be described. The second embodiment is the same as the first embodiment, except that step S4 in the flowchart (FIG. 3) corresponding to the first embodiment is improved. In the first embodiment, the file object name is stored in the character string storage buffer URLBUF from the access log file 15 in step S4, and the server computer 11 is accessed via the second proxy process 34. In the embodiment, access to a plurality of file objects is executed in an unordered manner and simultaneously in parallel in order to speed up processing.

In a TCP / IP network in which the present invention can be implemented, since communication is performed in units of packets, it is apparently possible to simultaneously communicate with a plurality of server computers in an external network. Therefore, a plurality of cache updater processes in step S4 can be performed simultaneously.

(2) Since a plurality of client computers 24 are connected to the internal network 23, the proxy transfer request of the file object by the proxy server device 53 is performed by intermingling access to a plurality of different server computers 11 on the external network 25. Therefore, since the access to the file object is performed through various routes on the external network 25, the actual transfer speed of the file object is determined by the bottleneck in the middle of the route.

One of the bottlenecks is the first communication path (the communication path 127 in FIG. 19) connecting the proxy server device 53 and the external network 25. Furthermore, there are various bottlenecks in the route to the plurality of server computers 11 on the external network 25. For this reason, the transfer speed from the external network 25 to the proxy server computer 53 is usually lower than the maximum transfer speed of this communication path. For example, even if the maximum transfer rate of the communication path (the communication path 127 in FIG. 19) connecting the proxy server device 53 and the external network 25 is 64 kbps, the transfer rate from the server computer overseas is one tenth or less. Often there are.

Therefore, it is possible to increase the transfer throughput by simultaneously performing a plurality of network connections so as to perform transfer close to the maximum transfer capacity of the first communication path 127 connecting the proxy server device 53 and the external network 25. That is, assuming that this network connection corresponds to one child process, a plurality of cache updater processes are started, and a plurality of network connections are performed in parallel so that the transfer capacity of the communication path 127 is used close to 100%. This speeds up the processing.

Therefore, the maximum number of parallel executions of the cache updater process is determined as the MAXPROCESS number so that a plurality of cache update accesses are executed simultaneously. This is a constant stored in the memory (the memory 206 in FIG. 19) in the flowchart described below. If this value is too large, management resources (memory) such as the OS process procedure are consumed, so it is necessary to determine the value depending on the operation state of the system. In the second embodiment, the number is set to 10.

Hereinafter, a specific processing procedure will be described with reference to FIG. First, a processes variable is defined in the memory 206 as a variable. In the background, the number of currently running cache updater processes is determined and assigned to the processes variable. The number of processes is obtained from the process management table of the OS (S421). Next, the processes variable is compared with the specified maximum number of child processes MAXPROCESS. If processes ≧ MAXPROCESS, the process moves to S423 and sleeps for a predetermined time (for example, 10 seconds). This is performed in order to wait for the process of the cache updater process running first to end (S423).

Next, it is determined whether or not the specified end time has come. If the specified time has passed, the cache update process ends the processing. If the specified time has not passed, the processing of S421 and subsequent steps is repeated (S424).

Also, if processes <MAXPROCESS in S423, by outputting an access request to the second proxy process 34, the second proxy process 34 starts one cache updater process as a background process (S425). In the UNIX (R) OS, the process can be executed in the background by starting the command line with an & symbol. Since the file object name acquired in S2 is stored in the character string buffer URLBUF, if the command shown in Expression 20 is executed, the cache updater process is executed as a background process.

Finally, it is determined whether or not the specified end time has been reached. If the specified time has passed, the cache update process ends the processing, and if the specified end time has not passed, the processing from S1 onward is repeated (S426).

処理 By the processing described above, simultaneous processing can be performed in a process less than MAXPROCESS.

According to the second embodiment, it is possible to prevent a delay in the cache update process from increasing by increasing the cache update process speed and always waiting for the cache update process. It should be noted that the processing in S2 in which the pattern file 30 is prepared and only the specific file object is to be updated is also effective.

As described above, in the second embodiment, since the cache update process is performed simultaneously, the time from the start to the end of the cache update is reduced, so that it follows the frequency of the access process from the user by the first proxy process 14. Then, the cache update processing can be advanced.

[Embodiment 3]
In the third embodiment, as in the second embodiment, the cache updater process is executed in parallel, but by further controlling the process, the user using the client computer 24 of the internal network 23 can execute the first process. Even if the cache update process 36 is executed during a time period when the server computer 11 of the external network 25 is frequently accessed using the proxy process 14, the number of cache update child processes is suppressed. In other words, a low priority is given to the cache updater process so that the user of the client computer 24 does not hinder the use of the network.

In addition, the network access of the user using the first proxy process 14 usually includes one page composed of a plurality of file objects (for example, an information page shown in FIG. 2). Access to a plurality of file objects occurs consecutively in time. Therefore, in the activation of the child process of the cache update access, by providing an idle time and delaying the activation of the child process, it is possible to increase the probability that the cache update access will occur after the user completes the page access.

Further, when the cache update access conditions for the same file object are met, that is, when the URLBUF of the first and second embodiments is not empty (when S3 in FIGS. 3 and 4 is yes), a fixed invalid time is provided. The cache update access to the server computer 11 is not performed within a predetermined time since the last time the cache was updated, so that when the client computer 24 frequently accesses the file object, the server update is performed by the cache update access. Is prevented from frequently occurring.

As a simple method of realizing this processing, the second proxy process 34 may be activated by setting the cache expiration date to an invalid time instead of 0. The invalid time may be 30 minutes, 1 hour, or the like, and may be about the minimum time of the average change frequency of the server computer 11. As a result, even if the cache update access is performed by the method of the first, second, or third embodiment within the invalid time, the cache will be hit, and the server computer 11 will not be accessed.

FIG. 6 is a flowchart showing the processing of the third embodiment. FIG. 6 is a flowchart showing the processing of the second embodiment, in which S421 and S425 in FIG. 5 are replaced with S431 and S435.

(4) Prior to performing the processing shown in FIG. 6, a constant called the maximum network connection number MAXPROCESS is determined. This constant is a constant indicating the maximum value obtained by adding the number of cache update child processes and the number of child processes activated by the first proxy process 14 for relaying network access. Therefore, the proxy server device 53 tries to restrict network access to the server computer 11 to MAXPROCESS or less. The processes in S1 to S3 in FIG. 6 are the same as those in the first and second embodiments, and thus description thereof will not be repeated.

(4) The number of cache updater processes running in the background and the number of network access relay processes of the first proxy process 14 are obtained and added to make a processes variable (Equation 21) (S431).

The $ processes variable is compared with the specified maximum child process MAXPROCESS, and if processes≥MAXPROCESS, the process proceeds to S433 and sleeps for a predetermined time (for example, 10 seconds). This waits until the processing of the number of cache updater processes started first is completed.

Next, it is determined whether or not the specified end time has been reached, and if the specified time has passed, the cache update process ends the processing. If the end time has not been reached, the process returns to S431 and continues (S434). If processes <MAXPROCESS in S432, one cache updater process accessed using the second proxy process 36 is used as a background process based on the extracted file object name (stored in the URLBUF). to start.

At this time, as shown in Expression 22, the child process sets the pause time to a DELAYTIME variable (unit: second), sleeps the OS pause function for DELAYTIME seconds using sleep, and then performs cache update access. When one page WWW data is composed of a plurality of image file objects as shown in FIG. 2, it is empirically known that a cache hit can be transferred from the proxy server device 53 to the client computer 24 in 10 seconds or less. Therefore, DELAYTIME is set to about 10 seconds.

Finally, it is determined whether or not the specified end time has been reached. If the specified time has passed, the cache update process ends the process. If the specified end time has not passed, the process returns to S1 and repeats the process (S436).

As described above, the cache updater process is controlled while limiting the total number of cache updater processes and the total number of network access relay processes of the user using the first proxy process 14. As a result, when the number of network accesses by the user using the first proxy process 14 increases, the cache updater process is less likely to be activated. Therefore, even if the cache update process 36 is executed during the time when the user is using the first proxy process 14, the cache updater process is given a low priority, thereby preventing the network activity of the user. Can not be.

Also, when the user accesses the file object in the cache, the transfer of one page normally ends within a few seconds. However, the cache updater process is started after a delay of DELAYTIME (seconds). If it is set to about 10 seconds, the user's file object access and cache update access are performed with a time lag (that is, pipeline), so that the load on the proxy server computer 53 is reduced and the user is provided with a comfortable network use. It is possible to do.

As described above, in the third embodiment, even when the cache update process 36 is executed during the time when the user is using the first proxy process 14, a low priority is given to the cache update child process. This has the effect of not hindering the user from using the network. Also, by introducing the delay time, the cache update process is performed after the user's continuous file object transfer process is completed, so the processing load on the proxy server computer 53 is reduced, and the file object transfer to the user is minimized. You can give priority. Further, since the invalid time is set, the client computer 24 accesses the same file object within the invalid time, and the cache update access hits the cache even if the cache update access conditions are met. No access to the server computer 11 occurs. As a result, even if the client computer 24 frequently accesses the same file object, time-consuming cache update access to the server computer 11 is suppressed, so that the load on the server computer 11 is reduced and the cache update process is made more efficient. I can do it.

[Embodiment 4]
The first to third embodiments relate to a proxy server computer. In the fourth embodiment, a gateway device (redirector computer) that performs only relay control of a file object will be described. FIG. 8 is a diagram showing a configuration in which cache server computers (proxy server computers) 65 to 67 and redirector computers 72 to 77 are connected via a local network 71.

The redirector computers 72 to 77 are one type of proxy server computers, but the redirector computers 72 to 77 according to the fourth embodiment only relay to the proxy server computers 65 to 67 and do not perform caching.

Hereinafter, the operation of the redirector computers 72 to 77 will be described in detail.

The redirector computers 72 to 77 are interposed between the plurality of proxy server computers 65 to 67 and the client computer group, are connected to the proxy server computers 65 to 67 via the local network 71, and are connected to the client computer group via the local network 78. Connected. The server computer 11 that provides the file object and the proxy server computers 65 to 67 are connected by the wide area network 61.

First, the redirector process in the redirector computer 73 receives the read request for the file object from the client computer 24 via the local network 78. The redirector computer can be realized by the workstations 200 to 400 shown in FIGS. In other words, a program corresponding to the redirector process is stored in the memory 206, and the function of the redirector computer is realized by the CPU 202 executing the program. The redirector process calculates the hash function based on the content of the received read request, and selects the proxy server computers 65 to 67 to relay according to the calculation result. Then, the redirector process relays the read request to the proxy process in the selected proxy server computer 65 to 67.

As an example to which the present invention can be applied very effectively, there is the WWW system on the Internet described above. FIG. 9 is a flowchart illustrating a processing procedure of the redirector computer on the WWW system.

In the WWW system, file object names distributed on a network are expressed and specified in a format called Uniform Resource Locator (URL).

{For example, assuming that the redirector computer 73 receives the read request from the client computer 24, the redirector computer 73 extracts the URL included in the request (S31). A calculation by a hash function is performed based on the extracted URL. Here, various hash functions can be considered. In the fourth embodiment, all the character codes of the characters included in the file object name expressed in the URL format are added, and the number of the proxy server computers 65 to 67 is calculated. The divided remainder was used (S33).

(4) The numbers of the proxy server computers 65 to 67 that relay the read request are determined based on the value calculated by the hash function (S34). Then, the read request is relayed to the selected proxy server computers 65 to 67 (S35).

From the viewpoint of effective use of the cache, it is preferable that all read requests for the same file object are relayed by the same proxy server computers 65 to 67. This is because if the same file object is redundantly stored in different proxy server computers, the cache capacity is reduced and the effective use of the cache file cannot be achieved. For that purpose, the hash function needs to be the same in the redirector processes in all the redirector computers 72 to 77.

In the above-described hash function, since the proxy server computers 65 to 67 are equally selected, the load on each of the proxy server computers 65 to 67 is equal. However, by appropriately selecting the hash function, each of the proxy server computers 65 to 67 is selected. It is also easy to intentionally distribute the load unevenly according to the capabilities of 65 to 67. That is, for a proxy server computer with a high processing capacity, the probability of being selected is increased, and for a proxy server computer with a low processing capacity, the probability of being selected is lowered, so that each proxy server computer has This is to equalize the processing speed.

(4) The proxy server computer indicated by the number calculated in this manner becomes the selected proxy server computer, and the redirector process relays the read request from the client computer 24 to the proxy process in the selected proxy server computer. The proxy process in the proxy server computers 65 to 67 relays the read request from the client computer 24 relayed by the redirector process in the redirector computer 73 to the server computer 11 via the wide area network 61, and transfers the transferred file object. Cache.

The redirector process can be implemented using a widely used DeleGate. The redirector process according to the fourth embodiment is realized by using the relay notation of the DeleGate and the filter function. The following is an example of relay notation.

In the expression 23, the first http indicates the name of the protocol to be used, and the proxyserver indicates the address of the proxy server computer on the network. Also, 10080 is called a TCP / IP connection port number and indicates to which process on the proxy server computer the read request is sent. -_- indicates that this is a DeleGate relay notation. The remaining http: // www. xxx. co. jp / test / index. html is the original read request URL.

When the DeleGate receives the above read request, it relays the read request of the file object indicated by the URL to the proxy process indicated by the port number 10080 of the proxy server computer proxyserver. That is, the proxy server computer to be relayed can be specified in the read request.

The filter function is a mechanism for extending the DeleteGate function using an external program. When the DeleGate receives a read request from a client computer or another proxy server computer, it temporarily passes the received read request to an external program, and receives the read request processed by the external program again to perform processing such as relaying. It is possible.

FIG. 11 is a diagram showing the internal configuration of the redirector computer. In the fourth embodiment, it is assumed that the redirector process 81 realized by the DeleGate receives a read request including the URL shown in Expression 24, for example.

This is www. xxx. co. jp indicates that this read request is transmitted to the server computer 11.

However, in the fourth embodiment, the proxy switching control unit 82 implemented as an external program calculates the hash function based on the URL (S32 to S33 in FIG. 9) by the filter function of the DeleteGate, and determines the proxy by the calculation result. Assuming that the server computer proxyserver1 is selected, the URL in the read request is rewritten as shown in Equation 25 (S34 in FIG. 9).

The redirector process 81 that has received the read request rewritten by the proxy switching control unit 82, which is an external program, sends the proxy server computer represented by proxyserver1 to www.proxyserver1. xxx. co. The read request to the server computer 11 represented by jp is relayed (S35 in FIG. 9).

{Accordingly, by changing the address of the proxy server computer in the URL to proxyserver1 based on the result calculated by the hash function, it is possible to select the proxy server computer to relay based on the URL of the original read request. Actually, two proxy server computers are prepared, and the hash function adds all the character codes of the characters of the file object names included in the URL (S32 in FIG. 8), and divides the remainder by dividing the number of proxy server computers by two. It was used (S33 in FIG. 8). That is, if the calculation result of the hash function is 0, relaying is performed by proxyserver0, and if it is 1, relaying is performed by proxyserver1.

FIG. 10 shows how the file objects of the same page shown in FIG. 2 are distributed to two proxy server computers. The upper part of FIG. 10 shows the URL of the original read request, the total character code of the file object name (check @ sum), the remainder obtained by dividing the total character code by 2 which is the number of proxy server computers, and the selected proxy server computer. Is shown. The lower part of FIG. 10 shows the URL rewritten by the proxy switching control unit 82. In this example, when the result of the hash function is 0, the number of relays to the proxy server 0 is increased. This is due to the small number of samples. In actual operation, it has been confirmed that an enormous number of accesses are evenly distributed approximately by half under the property that the distribution by the hash function is stochastically uniform.

As described above, when a plurality of proxy server computers are connected to the wide area network via different routes, distributing the access request from the client computer to the plurality of proxy server computers means distributing the request to the plurality of proxy server computers. But also. Therefore, the route can be effectively used, and the throughput is improved. Although the processing capacity of the proxy server computer is limited, access requests from the client computer are distributed to a plurality of proxy server computers, so that the number of access requests processed by one proxy server computer is reduced and throughput is reduced. improves.

{Circle over (2)} When a plurality of file objects are included in the same page, access requests to those file objects are also processed in a distributed manner by a plurality of proxy server computers, so that the access speed of a single page is improved.

Also, regardless of the relay request from any client computer, since the hash function for the same URL is the same, the same proxy server computer is always selected. If the same URL is accessed twice or more, the cache Access can be speeded up. Furthermore, since any function can be selected as long as the hash function is uniquely determined, any number of proxy server computers can be used.

In the fourth embodiment, it has been confirmed that the value obtained by adding all the character codes of the characters included in the URL used in the hash function (check @ sum) is about several thousands on average. For example, in the implementation at the Sharp Nara factory, the average of the total of characters and character codes included in the URL of the access request for 30,481 file objects measured on March 13, 1996 was 4438 (rounded to the decimal point). Was.

As described above, in the fourth embodiment, the remainder obtained by dividing the total number of characters and character codes included in the URL of the access request by the number of proxy server computers is used as a hash function, and is used for selecting the proxy server computer. . Therefore, although the number of the proxy server computers is two in the example of FIG. 10, even if the number is about several thousands, it is considered that the selection of the proxy server computers by the hash function is performed almost equally.

Also, if the hash function is the same, the number of redirector computers is arbitrary. Therefore, the scalability is very excellent in this method.

で は Also, in this method, the existing proxy server computer with cache can be used as it is. Therefore, the introduction is easier and more economical than the conventional method shown in FIG. Also, with regard to the client computer, it is only necessary to use the existing client computer and set the redirector computer as an existing proxy server computer, and it is not necessary to prepare a special client computer.

[Embodiment 5]
In the fifth embodiment, the redirector process executed on the independent computer in the fourth embodiment is executed on the client computer 24. FIG. 12 is a diagram showing the internal configuration of the device according to the fifth embodiment. A read request from a client process 84 (a process that operates in the same manner as the client computer 24 in FIG. 8) in the client computer 24 is transmitted on a local network. , And is directly transmitted to the redirector process 85 inside the client computer 24. This transmission is performed by, for example, UNIX (R) message communication.

The processing after the redirector process 85 receives the access request is the same as the redirector process 81 operating in the above-described redirector computers 72 to 77, and receives and processes an access request from another client computer via the local network. It is also possible. Therefore, detailed description will not be repeated here. Further, the function of the redirector process 85 can be directly incorporated into the program of the client process 84.

In the fifth embodiment, the communication between the client computer 24 and the redirector computers 72 to 77 shown in FIG. 8 is processed within the same client computer 24 in the fifth embodiment, so that the traffic on the local network can be reduced. Therefore, it is not necessary to prepare an independent redirector computer, and the function of the redirector computer can be realized by using the client computer 24, so that the cost can be reduced.

Embodiment 6
In the sixth embodiment, the redirector process executed on an independent computer in the fourth embodiment is executed on a proxy server computer. FIG. 13 is a diagram showing the internal configuration of the device according to the sixth embodiment. For example, when there is a read request from the client computer 24b, the read request is received by the redirector process 91 in the proxy server computer 99. . Thereafter, the redirector process 91 selects a proxy server computer that relays the access request as described with reference to FIG. 9, and if the selected proxy server computer is 98, the redirector process 91 sends the proxy process to the proxy process 86 of the proxy server computer 98. The access request is relayed (92). If both the proxy server computer on which the received redirector process is operating and the selected proxy server computer are 99, the read request is not transmitted over the network but is directly performed inside the proxy server computer 99 (93). ). Then, the proxy process (86 or 89) sends an access request to the server computer 11 via the wide area network 61 (62 or 63).

Similarly, when there is a read request from the client computer 24a, the read request is received by the redirector process 96 on the proxy server computer 98. Thereafter, the redirector process selects a proxy server computer, and if the selected proxy server computer is 99, relays an access request to the proxy process 89 of the proxy server computer 99 (95).

If the proxy server computer on which the received redirector process is operating and the selected proxy server computer are both 98, the read request is not sent out over the network but is directly transmitted inside the proxy server computer 98 (96 ). Then, the proxy process (86 or 89) sends an access request to the server computer 11 via the wide area network 61 (62 or 63).

The process of storing the file object acquired from the server computer 11 in the cache file (87 or 90) is the same as the above-described process, and therefore the detailed description will not be repeated.

{Also, it is possible to directly incorporate the function of the redirector process into a program that realizes the function as a proxy server on the proxy server computer.

Therefore, part of the communication between the redirector computer and the proxy server computer is processed inside the proxy server computer, so that the amount of communication on the local network can be reduced. In addition, there is no need to prepare an independent redirector computer, and the function of the redirector computer can be realized using the proxy server computer, so that the cost can be reduced.

Embodiment 7
The redirector computers 72 to 77 according to the fourth embodiment only relay access requests. However, in a large-scale local network in which the tree-like configuration described with reference to FIG. 23 is effective, by providing the cache function of the primary cache proxy server computers 503 to 508 to the redirector computer, And the load on the proxy server computer can be reduced. FIG. 14 is a diagram showing the system configuration. The redirector server computers 40 to 45 can be realized by combining the redirector computer 80 shown in FIG. 11 and the proxy server computer 13 shown in FIG. That is, since it can be realized by receiving the read request 17 of FIG. 16 by the redirector process 81 of FIG. 11, detailed description thereof will not be repeated. Also, by caching inside the client computer according to the fifth embodiment, it is possible to reduce the traffic on the local network and the load on the proxy server computer.

As described above, by caching in the redirector computer, the traffic on the local network and the load on the proxy server computer can be reduced.

Embodiment 8
Embodiments 4 to 7 described above do not have to be used exclusively. That is, in the eighth embodiment, the hash function used by the redirector function is the same in all the computers (proxy server computer, client computer, and redirector computer) that realize the redirector function, and all the access requests have the same proxy. Various computers described in the fourth to seventh embodiments can be used in a single network without any problem as long as the computers are relayed and cached by a computer having a cache function such as a server computer. Therefore, extremely flexible network operation can be performed in accordance with the operation circumstances of various networks and computers in each organization such as a company.

で は In a large-scale local network of a certain size or more, the connection to the wide area network may be made up of a plurality of routes (62 to 64 in FIG. 8). In such a case, if a different path is used for each proxy server computer, an access request from a client computer is distributed and relayed to a plurality of paths. Accordingly, the communication paths between the wide area network and the local network can be effectively used, and the throughput of communication between the wide area network and the local network can be improved.

に お い て In actual operation, it is not realistic to change the settings of all the computers at once. The transition usually takes a while. In order to achieve more efficient operation, it is necessary to make the hash functions identical in all computers having the redirector function. Therefore, either construct a system so that all the client computers use the redirector computer, or It is necessary to have a redirector function. However, it is possible to use a computer having a redirector function without constructing such a system. That is, in each system of the present invention, a part of the system can be provided with a redirector function without using a conventional proxy server computer or client computer at all, and the entire system can be used as it is.

Similarly, when the system is reconfigured, hash functions of a plurality of redirector functions may not match. Even in such a case, the operation of the entire system itself is not affected, so that it can be used as usual. Although use is possible in any of the above cases, it is not a desirable situation from the viewpoint of the proxy server computer and other effective use. Therefore, it is desirable to unify settings in computers having all the redirector functions as early as possible. However, even if the old and new settings are mixed during the transition period, the transition can be performed smoothly without causing unnecessary confusion.

The reference list including the http protocol is shown below.

It is an operation | movement conceptual diagram of the proxy server computer which is an example of the gateway apparatus in this invention. FIG. 11 is a diagram showing a display screen when an information page in a WWW server is read out to a client computer. 6 is a flowchart illustrating a processing procedure of a cache update process performed by the proxy server computer that is the gateway device according to the first embodiment. It is a flowchart which shows the processing procedure of the cache file management in a proxy process. 11 is a flowchart illustrating a processing procedure of a cache update process performed by a proxy server device that is a gateway device according to the second embodiment. 15 is a flowchart illustrating a processing procedure of a cache update process performed by a proxy server device that is a gateway device according to the third embodiment. 6 is a time chart of data transferred between a server computer and a client computer when the present invention is implemented. FIG. 14 is a diagram illustrating an entire configuration of a distributed file system according to a fourth embodiment. 15 is a flowchart illustrating a processing procedure of a redirector computer according to the fourth embodiment. FIG. 3 is a diagram showing how access requests to file objects of the same page shown in FIG. 2 are distributed to two server computers. FIG. 21 is a diagram illustrating an internal configuration of a redirector computer according to a fourth embodiment. FIG. 21 is a diagram illustrating an internal configuration of a client computer according to a fifth embodiment. FIG. 21 is a diagram illustrating an internal configuration of a proxy server computer according to a sixth embodiment. 146 is a diagram illustrating an entire configuration of a distributed file system in Embodiment 7. [FIG. FIG. 9 is a diagram showing a configuration of a conventional gateway computer. It is a conceptual diagram of a proxy server apparatus. FIG. 3 is a diagram illustrating a first example of a circuit configuration of a proxy server device. FIG. 9 is a diagram illustrating a second example of the circuit configuration of the proxy server device. FIG. 13 is a third example of the circuit configuration of the proxy server device, and is a diagram illustrating the circuit configuration of the proxy server device that is also used as the proxy server device in the embodiment of the present invention. 9 is a graph showing a relationship between a cache hit rate and an average access speed. 9 is a graph showing a relationship between a cache file size and a cache hit rate. 9 is a graph showing a relationship between a cache expiration date and the size of a cache file. FIG. 11 is a diagram illustrating a first example of the overall configuration of a conventional distributed file system. FIG. 11 is a diagram illustrating a second example of the overall configuration of a conventional distributed file system. FIG. 11 is a diagram illustrating a third example of the overall configuration of a conventional distributed file system. 9 is a time chart showing a relationship between updating data in a server computer and data read by a client computer in a conventional distributed file system.

Explanation of reference numerals

{11} server computer, 14} first proxy process, 15 access log, 16 cache file, 30 pattern file, 34 second proxy process, 35 FIFO buffer, 36 cache update process, 72-77 redirector computer.

Claims (9)

A client computer, a server computer for providing information in response to an information provision request from the client computer, a gateway device for controlling a file object of the server computer, and transfer of a file object based on the information provision request from the client computer And a proxy server computer that relays the information.
By determining a proxy server computer that stores requested information based on a calculation result of a predetermined calculation using a character string included in the information provision request and the number of proxy server computers in the system, each proxy server computer A gateway device for controlling the amount of information stored in the gateway. 2. The predetermined operation is a hash operation using a remainder obtained by adding a character code of a URL character string included in the information provision request and dividing the result by the number of proxy server computers in the system. Gateway device. The gateway according to claim 1 or 2, wherein the predetermined calculation uses the same calculation formula as another gateway device in the system, and the calculation result by the same information provision request determines the same proxy server computer. apparatus. A client computer, a server computer that provides information in response to an information provision request from the client computer, a control of a file object of the server computer, and a transfer of a file object transfer based on the information provision request from the client computer. A proxy server computer in a system consisting of a proxy server computer to perform.
By determining the proxy server computer that stores the requested information, including the own device, based on an operation result of a predetermined operation using the character string included in the information provision request and the number of the proxy server computers in the system. A proxy server computer for controlling the amount of information stored in each proxy server computer to be distributed. 5. The hash value calculation method according to claim 4, wherein the predetermined calculation is a hash calculation using a remainder obtained by adding a character code of a URL character string included in the information provision request and dividing the result by the number of proxy server computers in the system. Proxy server calculator. The said predetermined | prescribed operation | movement uses the same formula as another proxy server computer in a system, The operation result by the same information provision request determines the same proxy server computer, The Claims 4 or 5 characterized by the above-mentioned. proxy server computer. A client computer in a system comprising a client computer, a server computer that provides information in response to an information provision request from the client computer, and a proxy server computer that relays transfer of file objects based on the information provision request from the client computer And
The proxy server computer that stores the requested information is determined based on the result of a predetermined operation using the character string included in the information provision request and the number of the proxy server computers in the system. A client computer for controlling the amount of information stored in the client computer. 8. The hash calculation using a remainder obtained by adding a character code of a URL character string included in an information provision request and dividing the result by the number of proxy server computers in the system. Client calculator. 9. The client according to claim 7, wherein the predetermined calculation uses the same calculation formula as another client computer in the system, and a calculation result based on the same information provision request determines the same proxy server computer. calculator.

Images